BACKGROUND AND STUDY AIMS: Surveillance for patients with Barrett's esophagus is time consuming and subject to sampling error. Guided biopsies from neoplastic areas invisible on conventional endoscopy may increase effectiveness of surveillance examination. We assessed the accuracy of endocytoscopy in correlation with histology. METHODS: We analyzed 166 biopsy sites from 16 patients (13 male, mean age 62.1 years), without visible lesions, who presented for Barrett surveillance. Endocytoscopy images were recorded from pre-marked areas in the Barrett's segment using magnification x 1125 or x 450. Biopsies were taken from the same area to allow precise comparison with histology. Image sequences of each area were individually and blindly reviewed by a pathologist and a gastroenterologist. Major outcome variables included image quality, identification of neoplastic characteristics, and accuracy of endocytoscopy. RESULTS: Adenocarcinoma was histologically diagnosed in 4.2% of biopsy sites, high grade intraepithelial neoplasia (HGIN) in 16.9%, and low grade intraepithelial neoplasia (LGIN) in 12.1 %. Adequate assessment of endocytoscopy images was impossible in 49% of the pre-marked areas with magnification x 450 and in 22% with magnification x 1125. At most, 23% of images with lower magnification were interpretable to identify characteristics of neoplasia, and 41% with higher magnification. Interobserver agreement was fair at best (kappa from <0 to 0.45). Positive and negative predictive values for HGIN or cancer were 0.29 and 0.87, respectively, for magnification x 450 and 0.44 and 0.83, respectively, for magnification x 1125. CONCLUSION: When not supported by macroscopic evidence, endoscopic histology using endocytoscopy lacks sufficient image quality to be currently of assistance in identifying neoplastic areas.
